Gallup never releases their methodology, so it is always difficult to know what the internal poll statistics might imply. The overall sample shows high support for Obama relative to other polls (46%), so there's one clue.
The entire sample of Jewish respondents was 512 people, which to me implies a considerably larger margin of error than 7%, especially given the likely geographic concentration of those who answered the poll.
Also, because so many of those polled are likely to be life-long Democrats, any consideration of voting for a Republican would be far more likely to require the naming of an actual person, as opposed to a "hypothetical" character.
Sorry, Andy, if you read the end of the story, Gallup generally does release a summary of their methodology, which they did in this case. The reason why Gallup shows higher support for Obama than other polls is because they don't limit their respondents to registered voters, let alone likely voters. Instead, they interview anyone who answers the phone who claims to be 18 years or older. Heck, they even include aliens in most of their surveys.
The entire survey of Jewish respondents was 512 people, which to me implies a considerably larger margin of error than 7%.
I can tell you that there is a formula used in statistics where one can determine the margin of error to a specified probability (say 95%), based entirely on two variables which are plugged into that formula: the number of items sampled (in this case, 512 Jewish respondents) and the number of items in the total "universe" (in this case, the total number of Jewish Americans 18 and over, which is roughly 4 million). 7% happens to be a high margin of error for any national political poll, BTW. To lower the margin of error, they had to reach more Jewish Americans, but they probably decided not to because of the expense involved.
...the likely geographic concentration of those who answered the poll.
The Jews surveyed by their method would probably not have a geographic concentration much different than the general geographic distribution of Jews in the US. I don't see that as a particular problem of this methodology. But it's possible that the Orthodox and and more recently arrived Russian Jews - who tend to vote Republican in presidential elections and strongly anti-Obama - may have been undersampled because of cultural and language differences.
andy58-in-nh is correct about the unreliability of a 512 person "national" sample. The Presidential exit polls are comprised of over 17,000 respondents. In most results, you will find the Jewish results unlisted due to statistical insignificance, even in states like California, Illinois, even New York and Florida. If you have access to datasets with Congressional district info, the network sites like CNN seem to have pulled much of this, you'll find some results if you look to the urban areas with large Jewish populations. Of course you may also find results not much different than the general voting pattern of the district. Were the GOP interested in these voted, they'd be smart to find a formula to win over the liberal urban populations of the northeast, the Jewish vote would come right along.
The technique of tallying locally collected insignificant statistics to arrive at a "significant" statistic relative to the larger (national) group is highly suspect.
Self identified Jews, those who respond "Jewish" to the question of religion when asked are between 1.3% and 1.4% of the population. The National Jewish Population Survey gets to the 2% to 2.2% number by asking follow up questions to the self identified non-Jew, as to the Jewishness of family members or spouse. Which can get you counted as a Jew. As can having a personal relationship with Jesus/Jesus was a Jew.